Evaluating the impact of vitreomacular adhesion on anti-VEGF therapy for retinal vein occlusion using machine learning

Artificial intelligence in individualized retinal disease management

Owing to the rapid development of modern computer technologies, artificial intelligence (AI) has emerged as an essential instrument for intelligent analysis across a range of fields. AI has been proven to be highly effective in ophthalmology, where it is frequently used for identifying, diagnosing, and typing retinal diseases. An increasing number of researchers have begun to comprehensively map patients' retinal diseases using AI, which has made individualized clinical prediction and treatment possible. These include prognostic improvement, risk prediction, progression assessment, and interventional therapies for retinal diseases. Researchers have used a range of input data methods to increase the accuracy and dependability of the results, including the use of tabular, textual, or image-based input data. They also combined the analyses of multiple types of input data. To give ophthalmologists access to precise, individualized, and high-quality treatment strategies that will further optimize treatment outcomes, this review summarizes the latest findings in AI research related to the prediction and guidance of clinical diagnosis and treatment of retinal diseases.

Vitreoretinal Interface Abnormalities in Patients With Retinal Vein Occlusion in a Tertiary Referral Center

PURPOSE

The purpose of this study is to investigate the prevalence of vitreoretinal interface (VRI) disorders in patients with retinal vein occlusion (RVO) and to evaluate the impact of VRI abnormalities on the treatment outcomes of macular edema secondary to RVO using intravitreal aflibercept.

METHODS

Participants in this prospective study were consecutive patients with macular edema secondary to RVO, who received intravitreal aflibercept injections. At baseline, best-corrected visual acuity (BCVA) was assessed, and spectral domain-optical coherence tomography (SD-OCT) was performed to measure central subfield thickness (CST) and to evaluate the presence of VRI disorders, namely, vitreoretinal adhesion (VMA), vitreoretinal traction (VMT), epiretinal membrane (ERM), lamellar macular hole (LMH), and full-thickness macular hole (FTMH). The primary outcomes were the prevalence of various VRI disorders in patients with RVO and the impact of VRI disorders on BCVA and CST after aflibercept treatment in such patients.

RESULTS

At baseline, 16.1% of patients had VMA, 3.2% VMT, 18.3% ERM, and 1.1% LMH. There were a statistically significant improvement in BCVA and a decrease in CST in RVO patients over time. There was no statistically significant difference regarding BCVA and CST at baseline and until month 24 after treatment between patients with VRI disorders and those without VRI disorders. However, the mean number of injections during the follow-up period was higher in the group with VRI disorders (9.4±2.1) compared to those without VRI disorders (8.1±0.7, p=0.0002).

CONCLUSIONS

The prevalence of VRI disorders in patients with RVO was 16.1% for VMA, 3.2% for VMT, 18.3% for ERM, and 1.1% for LMH. VRI disorders were not found to affect the anatomical and visual outcomes after intravitreal aflibercept treatment in patients with RVO, although more intravitreal injections were needed in patients with VRI disorders.

Macular dynamics and visual acuity prognosis in retinal vein occlusions - ways to connect

Background and Objectives: This study aimed to establish possible connections between macular dynamics, various macular features, and visual acuity prognosis among patients with retinal vein occlusions. Materials and Methods: This study included 85 patients with central retinal vein occlusions (CRVO) and 26 with branch retinal vein occlusions (BRVO). We assessed macular features such as central macular thickness (CMT), foveal intraretinal hemorrhage (IRH), the presence and distribution of hyperreflective foci (HF), ellipsoid zone (EZ) disruption, inner retinal layer disorganization (DRIL), and posterior vitreous detachment (PVD), as well as their dynamics over one year of observation and their impact on final visual acuity prognosis, depending on the type of occlusion. Results: Best corrected visual acuity (BCVA) evolution is statistically significant regarding groups of age and type of occlusion and insignificant regarding gender. The best response to intravitreal treatment, quantified as a decrease in CMT, was registered after the first intravitreal injection. Connecting a decrease in CMT with BCVA improvement, we did not register a statistically significant correlation in the CRVO group, only in BRVO cases. The study results showed that complete PVD plays a significant positive role in decreasing CMT and BCVA improvement in cases of CRVO. Our study revealed that no matter the type of occlusion, the presence of foveal IRH will have a negative impact on the BCVA outcome. Statistically significant differences have been noted only for the evolution of visual acuity in non-ischemic CRVO cases, in correlation with the presence of EZ disruption. Outer retinal layer HF has proved to be a predictive factor for poor visual acuity outcomes. Conclusions: The most important non-imaging predicting factors regarding BCVA after retinal vein occlusions are age and baseline BCVA. CMT's dynamics still establish a weak connection with visual acuity fluctuations. The presence of foveal IRH, outer retinal layer HF, and foveal EZ disruption has a negative impact on visual acuity outcomes. Abbreviations: CRVO = central retinal vein occlusions, BRVO = branch retinal vein occlusions, CMT = central macular thickness, IRH = foveal intraretinal hemorrhage, HF = hyperreflective foci, EZ = ellipsoid zone disruption, DRIL = inner retinal layer disorganization, PVD = posterior vitreous detachment, BCVA = best corrected visual acuity, OCT = optical coherence tomography, BCVA Ti = best corrected visual acuity at first, BCVA Tf = best corrected visual acuity after one year, NR of IVI = number of intravitreal injections, SD = standard deviation, M = male, F = female, CMT Ti = central macular thickness at first, CMT T1 = central macular thickness after first injection, CMT T3 = central macular thickness after 3 injections, CMT Tf = central macular thickness after one year.

Efficacy and Safety of Ocriplasmin Use for Vitreomacular Adhesion and Its Predictive Factors: A Systematic Review and Meta-Analysis

Symptomatic vitreomacular adhesion (sVMA) impedes visual acuity and quality. Ocriplasmin is a recombinant protease, which may be injected into the vitreous cavity to treat this condition, yet controversy remains with respect to its effectiveness and safety, particularly its patient selection standard. In this systematic review, the PubMed, Embase, and the Cochrane Library were searched to identify studies published prior to August 2020 on the impact of ocriplasmin treatment on VMA release, macular hole (MH) closure, and/or related adverse events (AEs). Data were pooled using a random-effects model. Risk ratios (RRs) with 95% CIs were calculated. Of 1,186 articles reviewed, 5 randomized controlled trials and 50 cohort studies were ultimately included, representing 4,159 patients. Ocriplasmin significantly increased the rate of VMA release (RR, 3.61; 95% CI, 1.99–6.53; 28 days after treatment) and MH closure (RR, 3.84; 95% CI, 1.62–9.08; 28 days after treatment) and was associated with visual function improvement. No increased risk for overall AEs was seen in ocriplasmin treatment. The proportion of VMA release and MH closure in patients was 0.50 and 0.36, respectively. VMA release was more likely in patients with absence of epiretinal membrane (ERM). Patients with smaller MH diameter were more likely to achieve MH closure. Evidence from included studies suggests that ocriplasmin is a suitable and safe approach for treating sVMA. ERM and MH status are important factors when considering ocriplasmin treatment.

A Case for The Use of Artificial Intelligence in Glaucoma Assessment

We hypothesize that artificial intelligence applied to relevant clinical testing in glaucoma has the potential to enhance the ability to detect glaucoma. This premise was discussed at the recent Collaborative Community for Ophthalmic Imaging meeting, "The Future of Artificial Intelligence-Enabled Ophthalmic Image Interpretation: Accelerating Innovation and Implementation Pathways," held virtually September 3-4, 2020. The Collaborative Community in Ophthalmic Imaging (CCOI) is an independent self-governing consortium of stakeholders with broad international representation from academic institutions, government agencies, and the private sector whose mission is to act as a forum for the purpose of helping speed innovation in healthcare technology. It was one of the first two such organizations officially designated by the FDA in September 2019 in response to their announcement of the collaborative community program as a strategic priority for 2018-2020. Further information on the CCOI can be found online at their website (https://www.cc-oi.org/about). Artificial intelligence for glaucoma diagnosis would have high utility globally, as access to care is limited in many parts of the world and half of all people with glaucoma are unaware of their illness. The application of artificial intelligence technology to glaucoma diagnosis has the potential to broadly increase access to care worldwide, in essence flattening the Earth by providing expert level evaluation to individuals even in the most remote regions of the planet.

The prevalence of vitreomacular adhesion in eyes with macular oedema secondary to retinal vein occlusion selected for intravitreal injections

PURPOSE

To assess the prevalence of vitreomacular adhesion (VMA) in consecutive naïve eyes diagnosed with macular oedema (ME) secondary to retinal vein occlusion (RVO) and to longitudinally evaluate the incidence of vitreomacular interface changes over time and the influence on response to treatment.

DESIGN

Retrospective cross-sectional analysis and longitudinal cohort study conducted at two Italian tertiary referral centres.

METHODS

A total of 295 eyes, treated with intravitreal ranibizumab and/or dexamethasone for ME secondary to RVO between June 2008 and May 2018, were enrolled in the study. 280 fellow eyes met the inclusion criteria and were included as control group. The vitreomacular interface status was evaluated by spectral domain optical coherence tomography (OCT) and graded according to the OCT-based International Classification System developed by the International Vitreomacular Traction Study (IVTS) group.

RESULTS

At baseline, VMA was present in 130 (44.07%) RVO eyes and 142 (50.7%) control eyes (no statistically significant difference was found; p = 0.455). Mean follow-up (FU) was 35.98 months (min 6 - max 112). Throughout the FU, the incidence of spontaneous release of VMA (RVMA) in RVO eyes was significantly higher in comparison with that of the control group [59 (41.84%) RVO eyes versus 18 (12.33%) control eyes; p < 0.0001]. The number of injections in VMA+ eyes was significantly higher when compared with VMA- eyes. No significant difference was found between VMA+ and VMA- eyes regarding their mean best-corrected visual acuity (BCVA) at baseline and at each annual time point (p = 0.2). Differences in central macular thickness (CMT) were significant only at the baseline evaluation (p = 0.0303).

CONCLUSIONS

Vitreomacular adhesion (VMA) was not found to be more prevalent in eyes with RVO compared to healthy fellow eyes, and RVO, in turn, did not result in a higher persistence of VMA over time. This suggests that VMA and RVO might be two independent retinal phenomena, with no mutual pathogenetic influence. Vitreomacular adhesion (VMA) might have an impact on the response to treatment, since it was found to result in a more intensive treatment regimen; however, it did not affect visual and anatomic outcomes. These results do not support vitrectomy or PVD induction in the prevention, nor the treatment, of RVO.

Unraveling the deep learning gearbox in optical coherence tomography image segmentation towards explainable artificial intelligence

Machine learning has greatly facilitated the analysis of medical data, while the internal operations usually remain intransparent. To better comprehend these opaque procedures, a convolutional neural network for optical coherence tomography image segmentation was enhanced with a Traceable Relevance Explainability (T-REX) technique. The proposed application was based on three components: ground truth generation by multiple graders, calculation of Hamming distances among graders and the machine learning algorithm, as well as a smart data visualization ('neural recording'). An overall average variability of 1.75% between the human graders and the algorithm was found, slightly minor to 2.02% among human graders. The ambiguity in ground truth had noteworthy impact on machine learning results, which could be visualized. The convolutional neural network balanced between graders and allowed for modifiable predictions dependent on the compartment. Using the proposed T-REX setup, machine learning processes could be rendered more transparent and understandable, possibly leading to optimized applications.

Effect of posterior vitreous detachment on treat-and-extend versus monthly ranibizumab for neovascular age-related macular degeneration

AIMS

To investigate the impact of posterior vitreous detachment (PVD) on the efficacy of treat-and-extend (T&E) ranibizumab in neovascular age-related macular degeneration.

METHODS

In a post hoc analysis of a randomised controlled clinical trial, spectral-domain optical coherence tomography images of treatment-naïve patients randomised to receive T&E (n=265) or monthly (n=264) ranibizumab for 12 months were included. Certified, masked graders diagnosed the presence or the absence of complete PVD. The main outcome measures were the mean change in best-corrected visual acuity (BCVA) and central retinal thickness (CRT) at month 12, the number of administered ranibizumab injections and the proportion of patients extended to more than 8 weeks.

RESULTS

At baseline, complete PVD was present in 51% and 56% of patients in the monthly and T&E arms, respectively. Mean change in BCVA at month 12 was +9.0 (PVD) vs +9.5 letters (no PVD, p=0.78) in monthly treated eyes, and +6.0 (PVD) vs +7.5 letters (no PVD, p=0.42) in T&E treated eyes. Conversely, mean change in CRT at month 12 was -174 (PVD) vs -173 µm (no PVD, p=0.98) in the monthly arm, and -175 (PVD) vs -164 µm (no PVD, p=0.58) in the T&E arm. In T&E treated patients, the median number of injections was eight vs nine (p=0.035). 71% of PVD eyes were extended successfully, compared with 55% of eyes without PVD (p=0.005).

CONCLUSION

PVD was not found to impact functional and anatomical outcomes of T&E ranibizumab therapy. However, patients without a complete PVD required more retreatments and were significantly less likely to be successfully extended.

TRIAL REGISTRATION NUMBER

NCT01948830.

Guidelines for the Management of Retinal Vein Occlusion by the European Society of Retina Specialists (EURETINA)

The high prevalence of cardiovascular disease particularly in the elderly population is associated with retinal vascular disease. Retinal vein occlusions represent severe disturbances of the hypoxia-sensitive neurosensory retina. Acute and excessive leakage leads to the diagnostic hallmarks of retinal hemorrhage and edema with substantial retinal thickening. Advanced diagnostic tools such as OCT angiography allow to evaluate retinal ischemia and identify the risk for late complications and will soon reach clinical routine besides fluorescein angiography. Accordingly, the duration of non-perfusion is a crucial prognostic factor requiring timely therapeutic intervention. With immediate inhibition of vascular leakage, anti-VEGF substances excel as treatment of choice. Multiple clinical trials with optimal potential for functional benefit or a lesser regenerative spectrum have evaluated aflibercept, ranibizumab, and bevacizumab. As retinal vein occlusion is a chronic disease, long-term monitoring should be individualized to combine maintenance with practicability. While steroids may be considered in patients with systemic cardiovascular risk, surgery remains advisable only for very few patients. Destructive laser treatment is an option if reliable monitoring is not feasible. Ophthalmologists are also advised to perform a basic systemic workup to recognize systemic concomitants. The current edition of the EURETINA guidelines highlights the state-of-the-art recommendations based on the literature and expert opinions in retinal vein occlusion.

Automated OCT angiography image quality assessment using a deep learning algorithm

PURPOSE

To expedite and to standardize the process of image quality assessment in optical coherence tomography angiography (OCTA) using a specialized deep learning algorithm (DLA).

METHODS

Two hundred randomly chosen en-face macular OCTA images of the central 3 × 3 mm² superficial vascular plexus were evaluated retrospectively by an OCTA experienced reader. Images were defined either as sufficient (group 1, n = 100) or insufficient image quality (group 2, n = 100) based on Motion Artifact Score (MAS) and Segmentation Accuracy Score (SAS). Subsequently, a pre-trained multi-layer deep convolutional neural network (DCNN) was trained and validated with 160 of these en-face OCTA scans (group 1: 80; group 2: 80). Training accuracy, validation accuracy, and cross-entropy were computed. The DLA was tested in detecting 40 untrained OCTA images (group 1: 20; group 2: 20). An insufficient image quality probability score (IPS) and a sufficient image quality probability score (SPS) were calculated.

RESULTS

Training accuracy was 97%, validation accuracy 100%, and cross entropy 0.12. A total of 90% (18/20) of the OCTA images with insufficient image quality and 90% (18/20) with sufficient image quality were correctly classified by the DLA. Mean IPS was 0.88 ± 0.21, and mean SPS was 0.84 ± 0.19. Discrimination between both groups was highly significant (p < 0.001). Sensitivity of the DLA was 90.0%, specificity 90.0%, and accuracy 90.0%. Coefficients of variation were 0.96 ± 1.9% (insufficient quality) and 1.14 ± 1.6% (sufficient quality).

CONCLUSIONS

Deep learning (DL) appears to be a potential approach to automatically distinguish between sufficient and insufficient OCTA image quality. DL may contribute to establish image quality standards in this recent imaging modality.

Artificial intelligence in retina

Major advances in diagnostic technologies are offering unprecedented insight into the condition of the retina and beyond ocular disease. Digital images providing millions of morphological datasets can fast and non-invasively be analyzed in a comprehensive manner using artificial intelligence (AI). Methods based on machine learning (ML) and particularly deep learning (DL) are able to identify, localize and quantify pathological features in almost every macular and retinal disease. Convolutional neural networks thereby mimic the path of the human brain for object recognition through learning of pathological features from training sets, supervised ML, or even extrapolation from patterns recognized independently, unsupervised ML. The methods of AI-based retinal analyses are diverse and differ widely in their applicability, interpretability and reliability in different datasets and diseases. Fully automated AI-based systems have recently been approved for screening of diabetic retinopathy (DR). The overall potential of ML/DL includes screening, diagnostic grading as well as guidance of therapy with automated detection of disease activity, recurrences, quantification of therapeutic effects and identification of relevant targets for novel therapeutic approaches. Prediction and prognostic conclusions further expand the potential benefit of AI in retina which will enable personalized health care as well as large scale management and will empower the ophthalmologist to provide high quality diagnosis/therapy and successfully deal with the complexity of 21st century ophthalmology.

Automated detection of exudative age-related macular degeneration in spectral domain optical coherence tomography using deep learning

PURPOSE

Our purpose was to use deep learning for the automated detection of age-related macular degeneration (AMD) in spectral domain optical coherence tomography (SD-OCT).

METHODS

A total of 1112 cross-section SD-OCT images of patients with exudative AMD and a healthy control group were used for this study. In the first step, an open-source multi-layer deep convolutional neural network (DCNN), which was pretrained with 1.2 million images from ImageNet, was trained and validated with 1012 cross-section SD-OCT scans (AMD: 701; healthy: 311). During this procedure training accuracy, validation accuracy and cross-entropy were computed. The open-source deep learning framework TensorFlow™ (Google Inc., Mountain View, CA, USA) was used to accelerate the deep learning process. In the last step, a created DCNN classifier, using the information of the above mentioned deep learning process, was tested in detecting 100 untrained cross-section SD-OCT images (AMD: 50; healthy: 50). Therefore, an AMD testing score was computed: 0.98 or higher was presumed for AMD.

RESULTS

After an iteration of 500 training steps, the training accuracy and validation accuracies were 100%, and the cross-entropy was 0.005. The average AMD scores were 0.997 ± 0.003 in the AMD testing group and 0.9203 ± 0.085 in the healthy comparison group. The difference between the two groups was highly significant (p < 0.001).

CONCLUSIONS

With a deep learning-based approach using TensorFlow™, it is possible to detect AMD in SD-OCT with high sensitivity and specificity. With more image data, an expansion of this classifier for other macular diseases or further details in AMD is possible, suggesting an application for this model as a support in clinical decisions. Another possible future application would involve the individual prediction of the progress and success of therapy for different diseases by automatically detecting hidden image information.